Dynamic Programming Approach for Optimization of Approximate Decision Rules

We consider the notion of approximate decision rule, describe a way
for representation of the whole set of irredundant approximate decision rules
based on directed acyclic graph with nodes that are subtables of the
initial decision table, and present a way for sequential optimization
of approximate decision rules relative to the length, coverage, and number
of misclassifications. We present also results of experiments with decision
tables from UCI ML Repository.

Inhibitory Rules in Machine Learning and Data Mining
Mikhail Moshkov

We consider known results about possibilities to use inhibitory rules
in classifiers and for knowledge representation, and also consider new
algorithms for inhibitory rule optimization based on extensions of
dynamic programming.